Open-source security runtime for multi-agent AI. Enforces policy-based access control, sandboxes kernel interactions, and privacy-routes sensitive data for enterprise local inference.
Skills
Enforce policy-based access control at the agent boundary, restricting which tools and data each agent can reach
Enforce policy-based access control at the agent boundary, restricting which tools and data each agent can reach
Sandbox kernel-level interactions to prevent privilege escalation and contain runaway agents
Sandbox kernel-level interactions to prevent privilege escalation and contain runaway agents
Route sensitive data through privacy-preserving channels so PII never reaches untrusted model endpoints
Route sensitive data through privacy-preserving channels so PII never reaches untrusted model endpoints
Orchestrate trust levels across multi-agent hierarchies with per-agent permission scopes and audit trails
Orchestrate trust levels across multi-agent hierarchies with per-agent permission scopes and audit trails
Detect and alert on anomalous agent behavior patterns using real-time threat fingerprinting
Detect and alert on anomalous agent behavior patterns using real-time threat fingerprinting
Related Agents
NVIDIA OpenShell
Runs autonomous agents in isolated sandboxes with kernel-level policy enforcement over files, syscalls, and network, in…
AgentOps for Apify Builders Bundle
Apify actor bundle for agent builders: normalize run traces for QA, control costs, and guard tool calls with a firewall…
OpenSandbox
Runs AI-agent workloads in isolated Docker or Kubernetes sandboxes, exposing sandbox lifecycle, command, filesystem, an…
CubeSandbox
Runs self-hosted, KVM-isolated sandboxes for agent code execution behind an E2B-compatible API, with snapshots, cloning…